[EM] rbj's query re:DLW's assertion of Condorcet winners as ad hoc-ish due to political candidates being fuzzy options for voters.

[robert bristow-johnson]

On 10/31/11 12:32 PM, David L Wetzell wrote:
>
>
>     From: robert bristow-johnson <rbj at audioimagination.com
>     <mailto:rbj at audioimagination.com>>
>     To: election-methods at lists.electorama.com
>     <mailto:election-methods at lists.electorama.com>
>     Date: Mon, 31 Oct 2011 11:30:22 -0400
>     Subject: Re: [EM] hello from DLW of "A New Kind of Party":long
>     time electoral reform enthusiast/iconoclast-wannabe...
>     On 10/30/11 9:33 PM, David L Wetzell wrote:
>
>         So there's no cardinal or ordinal utility for any candidate
>         out there and all effective rankings of candidates used to
>         determine the Condorcet Candidate are ad hoc.
>
>
>     what do you mean by that?  do you mean that any of the
>     Condorcet-compliant methods are ad hoc?
>
>
> If my valuations of n candidates (or a subset therein) are fuzzy then 
> my rankings of them may be ad hoc and the odds are somebody's going to 
> be flipping a coin(possibly more than once) at some point.  Thus, 
> there's less of a benefit from being Condorcet-compliant than might 
> seem to be the case in theory.  As such, arguments that dismiss IRV or 
> IRV3 for failing to always be Condorcet-compliant are not slam dunks.

that is the case with any method.  here in the Chittenden state senate 
district, we elect 6 state senators out of maybe 15, where they are all 
running against each other regardless if they share party affiliation or 
not.  we're a pretty liberal place so 5 of the 6 elected turned out to 
be Dems with 1 GOP.  we *knew* that this one GOP would be reelected 
easily so then we knew that voting for one Dem might have the effect of 
voting against another Dem.  so then, for candidates that the voter knew 
little about or didn't have much of an opinion about, it was a 
crap-shoot about whether or not some voter would mark an X by their name 
or not.  clearly most voters did *not* exhaust all 6 of their votes.  i 
voted for 2 candidates that i particularly liked (and it turned out that 
my two votes sorta canceled each other since one candidate came in 6th 
and was elected, while the other came in 7th and was not).

so this is not a specific failing of Condorcet nor of the ranked-choice 
ballot in general.  at least with Condorcet, there is no logical problem 
with equal ranking of candidates on one's ballot (whereas there *is* a 
problem for IRV).  if you don't give a rat's ass about whom you prefer 
between Candidates C and D, rank them equally (if there are candidates 
that you know you prefer less).  if you don't like C and D at all, just 
don't rank them.

so if there is any ambiguity of a particular voter's preference between 
some group of candidates, that ambiguity is inherent with the voter, it 
belongs there on the ballot to reflect the voter's choice.  but even if 
the voter prefers Candidate C over Candidate D by *just* a teeny-weeny 
amount, just as with the case of the traditional ballot, it's 
"one-person one vote" and that voter's vote counts just as much as 
another voter who *really* prefers Candidate D over Candidate C a lot.  
their votes are equal because their franchise is equal (even if the 
relative strength of their preference is not equal).

this is why i don't like score voting except for Olympic competition.  
anyone reducing the "score" for some candidate they like from 10 (or 100 
or whatever the max) is volunteering that their vote does not carry 
equal weight with someone who is scoring their candidate up to the 
maximum.  i won't do that and i doubt many other voters will do that 
either.  if voters end up plugging the candidate of their choice up to 
10 and leaving the others at 0 (so as not to help these lessor 
candidates defeat their favorite), then score voting will devolve to 
plurality.

but i don't see why voter indecisiveness is a liability particular to 
Condorcet or to Ranked-Choice Voting in general.  by only asking "of the 
two candidates, who do you prefer more?" and not asking dumb 
quantitative questions like "how much more do you prefer one candidate 
over the other?", RCV extracts exactly the right amount of information 
from the voters.  Score demands too much (require ad-hoc evaluations), 
Approval not enough and, of course, the traditional ballot does not poll 
enough information from the voters when it's a serious 3-way (or more) race.

-- 

r b-j                  rbj at audioimagination.com

"Imagination is more important than knowledge."